A long optical cable loses photons. For a quantum connection, one proposed answer is to build the link in shorter pieces and join them. The catch is that those pieces rarely succeed at the same moment. Something has to remember the first success while the next attempt catches up.
An Innsbruck team has tested that idea using two trapped calcium atoms and a 17-kilometer stretch of existing municipal fiber. Its repeater node established entanglement between a photon sent across the city and another detected a few meters away in the laboratory. It is one relay with one long arm, not a completed quantum network.
The advance is where the cable went. An earlier version of this approach already worked with 50 kilometers of fiber wound into laboratory spools. The new experiment brings the link into a less controlled setting: rented fiber between a university laboratory and a suburban office.
Let the difficult side go first
Entanglement gives particles a shared quantum state, with correlations that ordinary independent descriptions cannot capture. A proposed repeater network would create shorter entangled links, then connect them through a process called entanglement swapping. This is not an ordinary amplifier copying a weak optical signal.
In the Innsbruck experiment, the researchers first made one calcium ion emit a photon entangled with it. That photon took the long route to the remote detector. Many attempts failed because the photon was lost or not detected.
After a successful detection, the ion retained its associated quantum state while the second ion tried to establish the short local link. The photon itself was not sitting in a memory: it had already been detected. The atom was doing the remembering.
Once both detections had succeeded, a joint operation and measurement on the two ions connected the correlations between the two photons. Measurements collected over many repetitions verified their entanglement. This was a test of the relay protocol, not a demonstration of two remote computers holding a usable quantum connection.
The order is practical. Start with the long, lossy link. Only then ask the first memory to wait while the easier local link is attempted. Memory turns two successes that need not happen together into a connected result.
What the city adds
The laboratory and office are about 5.2 kilometers apart in a straight line, but the cable follows a 17-kilometer route. The researchers rented two strands: one carried the quantum photons out, while the other returned ordinary control signals. Both were reserved for the experiment, rather than sharing traffic with Internet customers.
The ion's light also needed a change of wavelength to travel efficiently through telecom fiber. A converter shifted it into the band used by conventional fiber networks, while preserving the quantum information relevant to the experiment.
Outside the lab, connections and splices reduce transmission. The fiber's changing effect on light polarization can disturb the encoded state, and unwanted light adds noise. The team used filtering and periodic polarization corrections to keep the experiment working. Existing infrastructure was usable, but it was not maintenance-free.
A relay, not yet a service
The researchers recorded 3,369 successful two-photon events during a four-and-a-half-hour session. That session included calibration and setup changes. Their quoted rate of 0.69 events per second counts only the 81 minutes spent running the experimental sequence, not the whole session.
The result appears in a preprint posted on October 20. The authors say the underlying data will be released after peer-reviewed publication, so the result is not yet open to a separate raw-data reanalysis. The experiment also did not implement the final conditional correction that would turn its different measured outcomes into one standardized entangled state; that step was simulated numerically.
Connecting several such relays, keeping their memories reliable and maintaining useful rates are further tasks. But the city-fiber test answers a concrete question: this trapped-ion approach can handle more than a carefully arranged spool of cable. The useful trick is not making every photon arrive. It is remembering one success long enough to connect it with the next.
